Most AI you have used writes text. A System One model answers instead: give it a situation and a few typed questions (pick one option, score on a scale, yes or no), and it returns an answer for each, with a probability for every possible answer, in one fast pass. That makes it the right tool for the decisions inside software: routing a ticket, flagging a message, scoring a lead, checking a rule.
Getting started with System One models is a free course that takes you from that first idea to publishing a model of your own. Nine short lessons, about an hour in all, and nothing to install for the first half.
What the nine lessons cover
- What is a System One model? Typed questions in, probabilities out, and why honest probabilities matter.
- System One models and LLMs When a decision model beats a language model, and when it doesn't.
- How they decide How a state and its questions become one probability per answer.
- Your first API call An API key, two questions over HTTP and from Python, and reading the answers.
- Pull a model Download a model from the registry, files checked against their hashes.
- Run a model on your machine The same model on your laptop's CPU.
- Check, measure and calibrate with NoulXP Check that a model answers exactly as its authors' code does, and calibrate it to your own labels.
- Fine-tune on your own data Teach a model your own decisions.
- Publish your model Put it on the registry for others to use.
The early lessons run models in your browser, so you see a model decide before you write a line of code. The later ones give commands you can run on an ordinary laptop.
The certificate
Finish the nine lessons and claim a certificate of completion with your name on it. It is built to be checked, not just framed:
- Its own page. Every certificate lives at
systemonemodels.ai/certificates/followed by its number. The page says who it was issued to, when, and whether it is still valid. - A QR code. Printed on the certificate, it leads straight to that page, so anyone holding a printed copy can check it with a phone.
- LinkedIn in one click. The "Add to LinkedIn profile" button fills in LinkedIn's certification form for you: the course, System One Models as the issuer, the date, the credential number and its address.
- Copies to keep. Download it as an image, or print it and save it as a PDF from the certificate's page: one A4 page, ready to attach to an application.
Claiming it needs a free account with a verified email address, so every certificate belongs to an account with a working mailbox. Completion is self-reported: you mark each lesson done, there is no exam, and the certificate says so beside the account it was issued to.
Who it is for
Developers, machine learning engineers and product builders who have a decision in their code
that an if statement can't make and a chat model makes too slowly, too expensively or with too
little certainty about how sure it is. No machine learning background is needed to start; the
lessons on fine-tuning assume you are comfortable at a terminal.
Start the course, or read what a System One model is first.